There's a structural irony in how most businesses record their work: the people who generate the events — the salesperson closing an order in a customer's warehouse, the driver paying cash for a repair on the roadside, the engineer confirming a delivery on a site — are exactly the people furthest from a keyboard. So the event gets captured the only way field reality allows: a voice note to the office, a photo in a WhatsApp group, a promise to "enter it when I'm back."
Everyone knows what happens next. The voice note is heard, half-remembered, and typed up by someone who wasn't there — or it isn't, and Thursday's diesel receipt surfaces in a pocket the following month. The gap between when work happens and when it's recorded is where small businesses lose accuracy, and it's a gap that keyboards will never close, because the field is not a typing environment.
Push-to-talk closes the gap
BIZA's assistant has a push-to-talk mode: hold the mic, say what happened, release. In whatever language you speak:
- "Invoice Najd Logistics for today's delivery — 60 bags of cement at 18.50."
- "Record 350 fuel expense for the Hilux, paid cash."
- "Quotation for Marina Foods: 20 cartons of olive oil at 95 a carton, delivery included."
- "How much does Falcon Trading still owe us?"
The assistant does what it always does with a plain-language request: resolves the names against your real customers and products, applies the prices and tax treatment, and produces either an answer or a draft — a complete document you can read on the same phone screen, correct if needed, and submit.
The event gets captured at the moment it exists, by the person it happened to, in the words they'd have used to tell a colleague. That's the whole trick. Thirty seconds of speech at the tailgate of a truck replaces a mental note, a paper scrap, and a data-entry session that might never happen.
Why voice-first changes who can use the ERP
Something quieter happens when talking becomes an interface: the set of people who can operate the system stops being the set of people who learned it.
A driver who would never open an invoicing form can say what he spent. A veteran salesperson — brilliant with customers, allergic to software — can quote from the car in Arabic, and the system's record is as clean as the accountant's. Businesses in our launch markets run on exactly these mixed teams: different languages, different comfort with software, one set of books. Voice doesn't lower the bar for accuracy; it lowers the bar for participation, and accuracy follows because events are captured fresh instead of reconstructed.
The office's role shifts accordingly — from typing pool to review desk. Drafts arrive already structured; the accountant reviews and posts instead of transcribing. Same control point as always: nothing touches the books until a person approves it. A field draft with an ambiguous customer name waits, flagged, for a human decision — it doesn't guess its way into your ledger.
Voice, photos, and the capture habit
Push-to-talk pairs naturally with the camera. The spoken part carries intent — what happened — and the photographed part carries evidence: the supplier's invoice, the receipt, the delivery note. AI document capture reads the paper; the voice note supplies the context a document alone lacks ("this is for the Dammam site, split it across the two projects").
Together they form a field habit that actually sticks, because it's less work than the workaround, not more. Nobody sustains a habit of after-hours data entry. Everyone sustains a habit of talking.
What to expect honestly
Voice input is at its best on the routine and the concrete: known customers, known products, amounts and quantities. It's weakest where any interface is weak — genuinely ambiguous instructions, brand-new entities it has never seen, noisy environments that defeat the microphone. The design assumes this: unclear pieces come back as questions or flagged fields, and the draft-then-approve loop means a misheard "fifteen" versus "fifty" is a ten-second correction on a screen, not a wrong number in your books.
Start your team on expenses and enquiries — low-stakes, instantly useful — and let invoicing and quotations follow as trust builds. That's how the habit forms in practice.
How BIZA helps
Push-to-talk is part of BIZA's assistant on every device and every plan — any language; drafts for everything; approval always human. It runs on the same data, permissions, and audit trail as the rest of the ERP, so a record born from a voice note is indistinguishable from one typed at a desk — except that it actually got recorded.
See the AI assistant, or try it with your own voice.